Price Bubbles in Asset Markets
Abstract
From past to the present, many economies experienced price bubbles in asset markets and faced serious economic consequences since bubbles are inherently associated with a collapse, after a dramatic asset price increase. While policy makers still try to understand the main reasons of the bubbles, experiments may give great reasoning and may assist policy makers to implement various financial policies, in order to prevent or at least diminish the hazardous effects of the bubbles. As the global financial system has become more globally integrated since the 2008 global financial crisis and as the modern world today is much more deeply connected by capital flows and ideas today, it is, therefore, utmost of importance to understand the nature of the bubbles to foreseen next potential ones in advance. Although the main drivers of the bubbles can be associated to some certain reasons, still, human reaction/experience as another main driver is still a huge debate when determining the reasoning of the bubbles.
This paper is prepared to provide reasoning behind the bubbles and bring suggestions on how to better understand future bubbles by using experimental evidence. On that account, below research questions are aimed to be explained.
· What are the observable driven reasons which lead to bubbles in asset markets?
· Does the experience play a role in bubble formation? If not, why?
· What are the potential threats which can pose a risk for the success of experimental studies?
· How the experimental studies can help policy makers to prevent bubbles?
Price Bubbles in Asset Markets
1. Introduction
1.1. Definition of an Asset
An asset is basically a resource that provides some economic and useful value to its holder. While assets might refer to many sub-categories, it generally corresponds to any cash and equivalents, inventories, notes/bills, accounts receivables, securities, goodwill, fixtures, intellectual properties, machinery, real estate/lands etc.
1.2. Definition of a Bubble
A bubble is an economic cycle that is characterized by the rapid escalation of “normal market value or intrinsic value”, particularly in the price of assets. However, since price bubbles are sustained by expectations of future increases in the price of an asset, typically, it can be argued that a bubble is driven by or attributed to exuberant market/investor behaviour. However, the change of the behaviour is debated since empirical investigation is difficult.
While bubbles, in that sense, resembles rapid growth in markets and economies, it can be said that a bubble occurs any time when the price of a good rises far above the asset’s real value. At the end of a bubble, a quick decrease in value causes prices to deflate.
2. Reasons for a Price Bubble
Although efficient market hypothesis suggests that asset prices fully reflect all available information and they are always traded at their fair value on exchanges, making it impossible for investors to buy undervalued assets or sell them for inflated prices despite some fundamental or technical analysis, asset bubbles are evidence that prices can seriously deviate from their fair values.
Understanding the reasons which causes bubbles is actually widely argued by a host of economists. While it is hard to answer the reasons of a price bubble by indicating or providing some specific reasoning, it is known that bubbles are usually only identified and studied in retrospect, after a massive drop in prices occurs, in which make it even harder to understand the justification of the bubbles.
Although some economists even disagree that bubbles occur at all, on the basis that asset prices frequently deviate from their intrinsic value, there are also historical events which show devastating consequences of price bubbles on economies. Therefore, it could be beneficial to briefly explain these well-known events in order and try to understand why these bubbles cannot perfectly be explained.
2.1. Japanese Land Bubble (Late 1980s)
After partially deregulation of Japanese Banks, Japanese economy experienced a huge surge in the prices of real estate. While deregulation eased the access to credit, speculation encouraged by the government’s fiscal policies (low tax rates, loophole-ridden tax laws on capital gains from land, etc.) led many investors to try making profit from the current status-quo. Plus, it was known that the real estate/land supply was very limited in Japan. Therefore, while valuation expectations were always higher, bubble burst occurred inevitably.
2.2. Dot-com Bubble (Late 1990s)
The dot-com bubble was characterized by excessive speculation in internet-related companies in the U.S. People bought technology stocks at high prices and believed that they could sell them at a higher price since in that era, rapid technology growth convinced people that these tech-stocks were an opportunity and a reasonable way to make a profit. Nonetheless, after people’s confidence on these stocks and companies was lost, a large market correction occurred.
2.3. Mortgage Bubble (Late 2000s)
Starting reason of Mortgage Bubble in the U.S. was indeed unprecedented growth of the subprime mortgages. U.S. government-sponsored mortgage lenders made home loans accessible to borrowers who had low credit scores and a higher risk of defaulting on loans. While this program was firstly initiated to solve the vast housing problem in the United States;
· Adverse selection led many people to have unaffordable mortgages, as well as belief that prices would rise up.
· Investment vehicles creatively derived from mortgage market by financial institutions could not anticipate explosion of consumer debt. Accordingly, the financial turmoil of 2008 occurred, and stock market crashed.
2.4. Deemed Observable Results
· Low interest rates: As a matter of fact, low interest rates make it easy to have inexpensive credits for people. While this allows people to spend more, the greater spending power results in prices rising due to increased demand.
· Demand-pull inflation: The greater demand for an asset leads to a price boost for the asset. Nonetheless, the price increase may be seen as an indicator of future increase of the price. This may lead to the formation of a speculative bubble.
· Supply shortage: The reduced supply or the expectation of a reduction in the supply of an asset in the future may lead to a greater demand for the assets. Hence, investors may think that there are only a limited number of assets available in the market, and they rush to buy as much as possible.
3. Understanding Bubbles with Experiments
Other than observable results for the bubbles, investor behaviour is also one of the key drivers in the reasoning of bubble formation. Normally a price bubble illustrates an unstable equilibrium, in which further indicates that forces of supply and demand in the market do not correct price deviations away from the equilibrium price. Therefore, to understand the price deviations, experiments can be utilized. However, design of the experiment matters.
3.1. Experimental Design
As per the study conducted by Anita Kopányi-Peuker and Matthias Weber (2020) to study the role of investor experience in the formation of asset price bubbles, two separate experiments (i. experiment which investors actually trade, ii. experiment which investors only forecast future prices) were designed. To make the experiments comparable, below standards were accepted as rules;
· A sufficient number of periods to be able to observe bubbles was given.
· Each multiperiod market was repeated three times. While the investors did not know the number of periods per round, they knew that periods were between 25 and 40. The rounds were otherwise identical.
· Only one round was paid out. The round for payment was randomly selected by the computer at the individual level.
· Interest rates and dividend processes were identical. Consequently, the fundamental values were identical.
· Investors were randomized to groups of six and the group composition remained fixed throughout the experiments.
· Market price in period t depended on the expectations of the price in period t +1.
· An upper limit was standardized.
Since the aim of this experiment was to understand the impact of investor behaviour, information about the fair price of the asset given to investors prior to beginning of the experiments was differentiated. Yet, investors always had full information about the interest rate and the dividend process, so that they were able to always calculate the fundamental value of the asset.
Accordingly, three information treatment had been formed.
i. NO-INFO treatments: Investors received no explicit information about the buyout price.
ii. INFO-AFTER treatments: Buyout price was communicated after the round ends. Although the fundamental value was the same in the different rounds, information was repeated.
iii. FULL-INFO treatments: Buyout price was communicated before the experiment started.
3.2. The Call Market Experiment
Participants traded assets with each other. Like in the real-life stock exchange, initial cash accounts were used to buy assets and each investor interacted with others throughout the experiment. Therefore, this experiment resembles double auction experiments. Investors simultaneously placed ask or bid options in the market and the computer calculated the aggregate demand and supply schedules. In this way, market price was determined. Nonetheless, it should be noted that in this experiment;
· Investors were not be able to make short selling. However, speculation was not restricted. Investors traded with each other in possibility of the realization of capital gains.
· Investors were not be able to enter bids that they would not be able to pay for with the points in their cash account.
· Investors had the possibility to skip periods when they did not want to trade. When this was the case, computer automatically proceeded to the next period. Because of that, this experiment was open to active participation hypothesis, which indicated in this experiment that participants as investors may have still felt that they were supposed to buy and sell the asset since they were placed into this environmental design with the role of an investor and the buying and selling was the experiment’s objective.
· Ask prices had to be higher than bid prices. Investors could not buy assets from themselves.
· After the trade in a period, dividend and interest earnings were paid. Both dividend and interest earnings were paid to a separate savings account, which yielded interest but could not be used for buying assets. Therefore, the cash-to-asset ratio was constant over time. It is here important to highlight that higher cash-to-asset ratios may lead to higher asset prices with inexperienced subjects according to some other studies.
3.3. The Learning-to-Forecast Experiment
Participants only forecasted future prices and these forecasts were computerized. Therefore, there was no actual trade, yet the predictions of the investors. Investors behaved like advisors and tried to predict the future price of the asset. Computerized companies traded based on the advisers’ forecasts and the market price was determined.
Companies tried to maximize their utility by deciding how many assets to hold. Since investors as advisors gains merely depended on their forecasting accuracy, advisors were supposed to predict future prices as accurately as possible. Nonetheless, it should be noted that in this experiment;
· Advisors were aware of prediction upper limit.
· Likewise in the call-market experiment, investors had the possibility to not decide a forecast in a given time. However, when a forecast was not submitted, corresponding company remained inactive, and the advisor earned no points for the given period. Therefore, here the effect of active participation hypothesis was limited.
· Advisors knew that the price positively depended on the submitted forecasts.
3.4. Results of the Experiments
It is phenomena that bubbles are formed due to inexperienced investors. Therefore, related authorities who are responsible for financial stability maybe should not be too concerned about price movements in markets with mainly professional investors, such as the stock market.
However, results do not show evidence for above-mentioned phenomena. Outcomes are argued in below.
3.4.1. Key Findings in a Nutshell
Although previously definition of a bubble is given, to determine it in more specific terms, bubbles in these experiments were precisely defined when the average price across all periods of a round was at least twice the fundamental value. On that account, it was observed that:
· When taking all the prices into account and calculate a mean price for the all groups and rounds, it was noticed that bubbles indeed do not disappear with experience in both call market experiment and learning-to-forecast experiment, since the mean prices were considerably above the fundamental value (more than twice). Moreover, by discouraging the expectations, it was observed that information level (INFO-AFTER and FULL-INFO) of the investors actually played a little impact on the improvement of price levels since only a very slight and hardly visible downward trend was observed with the acknowledged investors. Therefore, it was clearly resulted that bubbles do not disappear, and they are independent of the amount of information that investors receive about the fundamental value of the asset.
· It is also an interesting fact that shape of the bubbles on these experiments differed. By saying shapes, actually time-duration of a bubble can be interpreted. On the call market experiment, long periods of severe mispricing had been observed and market price did not change a lot from period to period. Accordingly, it was such bubbles defined as flat bubbles. However, as mentioned before, bubbles can burst and deflate in time. While a burst indicates a fast change of prices, deflation indicates a slow decrease of market prices toward the fundamental value. In that case, flatness of the bubble differs. Plus, without showing prior signs of bursting or deflating, there also could be sustained flat bubbles which last until the market is terminated. In flat bubbles, relatively low standard deviations of prices had been observed since the consecutive prices were similar. It is also argued that some investors may have expected the market to continue for longer and therefore higher prices were traded. It is here important to mention that there were not many bubbles that do not burst or deflate before the market ends in the later rounds of INFO-AFTER and FULL-INFO groups.
· In the learning-to-forecast experiment, nonetheless, boom-and-bust cycles had been observed always. Prices rised and decreased smoothly, but with large amplitudes. Therefore, these bubbles had high standard deviations since there were a host of very different prices.
· As per the given bubble shape formation between experiments, the main question lies to find out why the bubble shapes differed. It was argued that since the investors in the call market experiment had the opportunity to not participate trading, it was ambiguous to detect differences of the investors’ strategies during the times of no trade. Furthermore, investors in the learning-to-forecast experiment were actually advisors and their decisions were strategic complements. Therefore, in times when investors expected other investors to submit high forecasts, their best decision was to also submit high forecasts. Since it was known that the gain of advisors depended on the accuracy of the estimation, others’ forecast may have created an impact on others’ decision. Plus, adaptive expectations played a role here too. However, it was not sufficient to explain large and rapidly forming bubbles.
· What can be derived from the behavioural difference of the investors can be the terminology of incentives, too. It is because, while investors in call market experiments traded assets, there were no such clear incentives to complement others’ actions. These investors simply purchased an asset, expected higher prices, and hoped to sell their assets in higher prices.
· As previously said, pricing of the assets hardly improved over time. However, it had been observed that bubbles appeared earlier in the later rounds of both experiments, rather than in the first rounds. Although there was no established measure for how early in a market a bubble occurs, it was stated that since the pricing behaviour of the investors change over time, bubbles seemed to speed up in later stages in both experiments.
· While the results of learning-to-forecast experiment were novelty, results of call market experiment have some similar outcomes with the Hussam, Porter, and Smith (2008) study which indicates that experience alone is not a sufficient condition to ensure the elimination of price bubbles and it is still possible to rekindle bubbles with experienced subjects.
3.4.2. Additional Findings for Call Market Experiment
Although it has been stated in the section 3.2. that the cash-to-asset ratio was constant over time, after carrying-out these experiments, a new treatment with a low cash-to-asset ratio was also implemented. Notably, it was reached that there were considerably fewer and smaller bubbles in the call markets. In the majority of groups, pricing was improved, and it was relatively close to the fundamental value. Especially with the FULL-INFO group with lower cash-to-asset ratio, some learning was even achieved.
On the other hand, since the cash constraints were absent by the design in the learning-to-forecast experiment, there were no impact of cash-to-asset ratio. Therefore, level of this ratio broke the similarities (in terms results) of the experiments.
3.4.3. Additional Findings for Learning-to-Forecast Experiment
After the experiments, it was decided to implement an additional model for the learning-to-forecast group. While the model was depending on reasoning level, investors who had higher level of reasoning reacted to the actions of the ones who had lower level of reasoning. Although, it was observed that low level investors acted more sophisticated in time, however, again large, and rapidly rising bubbles in the earlier in later rounds of the experiment had been observed. It was argued that higher levels of reasoning led to stronger price bubbles because of strong positive feedback from price expectations to realized prices in the market.
4. Potential Threats to Further Studies
It would be fair to expect that all investors would have rational expectations for their investments. However, in today’s instantly changing world, the interaction between rational and behavioural traders is a deeply argued issue. Since the human interaction has increased considerably as the technology advances in time, investors, especially individual investors are enabled to get and share information faster. Below, some examples which poses a challenge for experimental economics are given;
· In 2021, a short squeeze incident occurred in U.S. stock exchanges. Gathered investors on Reddit application has targeted big investment corporations which do heavily short operations in some stocks, gains huge benefits and make loss for the small investors. In one night, many small investors purchased pre-determined hugely shorted stocks and forced big investment corporations to buy these stocks again to cover their loss situation. This led to even higher prices for these stocks and for a little moment, a price bubble was formed. Although the prices later dropped considerably, millions of dollars were written as loss for the investment companies. While still short squeezes continue randomly as of today, randomized human behaviour jeopardizes success of experimental studies.
· Additionally, cryptocurrency market where there are many inexperienced investors is fundamentally different and prone to bubbles, as well. Similar to short squeezing, when mainly inexperienced traders who are open to misleading and strategic misrepresentation are prone to mispricing, an inflow of them into the markets may increase the chance of bubble formations. Here, Twitter-made popular meme cryptos or tokens could be an example though. As Tuckett and Taffler (2008) argued that holding and selling assets in an unknown ambiguous environment leads to an integration of emotional experiences to behaviour, bubbles arise in markets for exotic/unknown assets. Since many cryptocurrencies and tokens are today unknown assets for many investors who actively trade, therefore, this craze makes the predictability of bubbles difficult and empirical studies troublesome to implement.
In summary of these two examples, it can be said that as per the inherent human decision-making, sometimes people may fail to take the rational choices usually assumed in the standard academic studies. Therefore, investor irrationality poses a risk.
Furthermore, it is also important to mention that bubbles can emerge if investors hold heterogenous beliefs, potentially due to psychological biases. Investors may agree to disagree about the fundamental value. Plus, it is fact that endowment effect is deeply rooted in people’s assessment of risk when the prices change. Therefore, while the experimental studies are useful to isolate, distinguish and test the validity of different designs, these unpredictable endogenous variables may not be easy to handle.
While it was also mentioned in the section 3.4.2. that low cash-to-asset ratio improved pricing, on the macro-economic level, however, it could be an indicator that above-mentioned occasions has specifically raised after the Covid-19 pandemic, due to response of many central banks’ excessive money supplying actions. When considering that nominal interest rates were also reduced in many countries, easier way to reach money made an enormous amount of excess liquidity in the global system. While arguing that the money has to go somewhere, it is possible that investors could have been unintentionally encouraged to above mentioned occasions. Plus, by growing empirical evidence, some studies indicate that holding interest rates too low for too long is likely to lead declining lending standards and growing risk-taking in the banking system, which may lead to bubbles and defaults over time.
5. Before Closing Suggestions for Policy Makers
Some economic historians such as Charles Kindleberger and others, have provided evidence that financial systems have a tendency to generate such financial boom-bust cycles, which sometimes take systemic dimensions that can cause or contribute to severe financial crises and recessions, which is an explosive path. In that respect, it is known that bubbles have long been intrigued by economists and several strands of models, empirical tests, and experimental studies still continue for the assurance of financial stability.
Even though the world recovered from 2008 global financial crises and individual banks looks healthier than ever, global financial system as a whole is more fragile than ever because of vast globalization effects. Therefore, in a more complex and interdependent system, when also considering that price bubbles hard to notice in advance, it may be a suggestion to implement an alarm mechanism for the different types of assets in different markets. In that extent, if current prices of an asset exceed long-term averages by a sufficiently large margin, a warning signal could be recorded and a specific experimental study to analyse a potential a bubble could be organized in post-haste. Therefore, to maintain financial stability, rather than trying to perceive next unknown bubble formation in the long run, markets can be monitored in best efforts with an alarm system and the risks can be prioritised in the short and medium run. Although this may not be cure, financial stability may be sustained in some degree.
Last but not least, as it has been argued on the experiments that bubbles occur whether the investors are experienced or not, it may be very valuable to carry out further implementation of these experiments with new variables, for the long run bubble detection studies. For example, while the discovery of the effect of cash-to-asset ratio indeed provided helpful insights, other important characteristics which play a role on investors’ asset purchase decision can be added to the experimental design. In that sense, different capital gains taxation mechanisms on the call market experiment or higher capital and liquidity limitations for the companies in the learning-to-forecast experiment may perhaps provide some other clues for the bubble formation. Although it is still a long way to understand the bubbles as a whole, experimental studies are the key to fully understand this hale and hearty organism, economy.
Discussed Topic
Anita Kopányi-Peuker and Matthias Weber (2020). Experience Does Not Eliminate Bubbles: Experimental Evidence. The Review of Financial Studies. vol 34 pp. 4450–4485
Other References
Hussam, Porter, and Smith. (2008). Can bubbles be rekindled with experienced subjects?. American Economic Review. vol 98 pp. 924–937.
Lei, Noussair and Plott (2001). Nonspeculative Bubbles in Experimental Asset Markets: Lack of Common Knowledge of Rationality vs. Actual Irrationality. Econometrica vol. 69, pp. 831–859.
Jingchi Liao, Cameron Peng, Ning Zhu (2021). Extrapolative Bubbles and Trading Volume.
The Review of Financial Studies. pp. 1–41.
Chernomas and Hudson (2016). Finance, Crisis and the Efficient Market Hypothesis. Pluto Press. pp. 125–146
Tumpel-Gugerell (2011). Asset price bubbles: how they build up and how to prevent them?. European Central Bank.
